opencorp

Find your users while you build. Drop in a link to what you're building — opencorp runs a research agent that finds your competitors and surfaces the Hacker News threads where your future users are already talking.

---

Demo

opencorp-demo.mp4

What it does

opencorp is an autonomous user-acquisition platform. Drop in a link to what you're building, and it runs a research agent that returns two things:

• Competitors — who else is building something similar, with the sources behind them
• Hacker News threads — where your future users are already talking, ranked by how relevant each thread is to your product

The agent picks its own queries, decides how many searches to run, and re-ranks results. You stay in your editor.

Features

• Product analysis — autonomously reads a product's site to extract name, description, features, audience, and pricing
• Competitor discovery — searches the web for adjacent tools, with each result traced back to a source
• HN thread discovery — surfaces Show HN launches, Ask HN threads, and discussions where the user can pitch
• Streaming UI — research results stream to the dashboard in real time, so the user sees tool calls and partial results as they happen
• Auth + persistence — per-user research sessions, full history (each session row stores product analysis, competitors, and HN threads)
• Multi-source research — combines web search, Hacker News, and the product's own site for grounded context

---

How it works

Product URL
    │
    ▼
Product Analyst ───► Reads the product's site
    │                  (homepage, pricing, docs, about, ...)
    ▼
┌──────────────────────┐
│  Discovery Agent     │  Picks queries, decides how many to run,
│  (autonomous)        │  re-ranks results
└──────────────────────┘
    │
    ├──► Competitor research task ──► List of competitors with sources
    │
    └──► HN threads task ──► Curated HN threads with relevance reasons
                                      │
                                      ▼
                              Dashboard UI (live stream)

---

Tech stack

• Framework: Next.js 16 (App Router) on React 19
• AI: Mastra agents + workflows, OpenRouter for LLM routing
• Tools: Exa (web search), HN Algolia API, Jina Reader (page fetch → markdown)
• Jobs: Trigger.dev v4 for long-running async research
• Auth & DB: Supabase (Postgres + RLS)
• Storage: LibSQL (default), DuckDB (observability)
• UI: shadcn/ui, Tailwind CSS v4, Motion, lucide-react, Geist
• Language: TypeScript, Zod v4

---

Quick start

Prerequisites

• Node.js 20+
• pnpm (this project uses pnpm exclusively — pnpm-workspace.yaml blocks other package managers' lifecycle scripts)
• Accounts + API keys for: OpenRouter, Exa, Jina, Supabase, Trigger.dev

Install

git clone https://github.com/KrishavRajSingh/opencorp.git
cd opencorp
pnpm install

Configure environment

Copy and fill in .env:

cp .env.example .env

See Environment variables below for the full list.

Run the database migrations

Open the Supabase SQL editor for your project and run the migrations in order:

1. supabase/migrations/0001_research_sessions.sql — creates the research_sessions table and its RLS policies
2. supabase/migrations/0002_hn_threads_result.sql — adds the hn_threads_result column for persisting curated HN thread output

See supabase/README.md for details.

Start the dev server

pnpm dev

The app runs at http://localhost:3000.

In a separate terminal, start the Trigger.dev worker (optional, only needed for background research tasks):

pnpm dlx trigger.dev dev

---

Environment variables

Variable	Required	Purpose
OPENROUTER_API_KEY	yes	LLM provider — Mastra routes all model calls through OpenRouter
JINA_API_KEY	yes	Used by fetchPageTool to read product pages as clean markdown
EXA_API_KEY	yes	Used by searchWebTool for semantic web search
TRIGGER_SECRET_KEY	yes	Authenticates the Next.js app → Trigger.dev worker
NEXT_PUBLIC_SUPABASE_URL	yes	Supabase project URL (browser-safe)
NEXT_PUBLIC_SUPABASE_PUBLISHABLE_KEY	yes	Supabase anon key (browser-safe)
SUPABASE_SECRET_KEY	yes	Supabase service-role key (server-only)
NEXT_PUBLIC_SITE_URL	yes	Public origin used for OAuth redirects and email links

.env is git-ignored. Never commit secrets.

---

Project structure

opencorp/
├── src/
│   ├── app/                       # Next.js App Router
│   │   ├── page.tsx               # Landing page
│   │   ├── auth/                  # Sign-in / sign-up flows
│   │   ├── dashboard/             # Research dashboard (projects, sessions)
│   │   └── api/research/          # Streaming research API + task triggers + persistence
│   ├── components/
│   │   ├── ui/                    # shadcn/ui primitives
│   │   ├── ai-elements/           # Custom AI chat / artifact UI
│   │   └── dashboard/             # Dashboard-specific blocks (HN threads, etc.)
│   ├── mastra/                    # Mastra AI framework
│   │   ├── agents/                # discovery, product-analyst, weather
│   │   ├── tools/                 # search-web, search-hn, fetch-page, weather
│   │   └── workflows/             # weather-workflow (example)
│   ├── trigger/                   # Trigger.dev task definitions
│   │   ├── research.ts            # product-research, competitor-research, hn-threads
│   │   └── streams.ts             # Real-time event streaming
│   └── lib/                       # Shared utilities (supabase clients, cn, etc.)
├── supabase/
│   └── migrations/                # SQL migrations (run in order)
├── trigger.config.ts              # Trigger.dev project config
├── next.config.ts
├── components.json                # shadcn/ui config
└── pnpm-workspace.yaml

---

Scripts

Command	What it does
pnpm dev	Start the Next.js dev server with HMR
pnpm build	Production build
pnpm start	Run the production build
pnpm lint	ESLint (no typecheck script is configured — run tsc --noEmit if needed)

---

Architecture notes

• Streaming research — research results stream to the dashboard in real time as the agent works, not after it finishes.
• One agent, three tasks — productAnalystAgent runs alone for product analysis; discoveryAgent powers both the competitor-research task and the hn-threads task. Each task writes its structured output back into its own column on the research_sessions row.
• Mastra storage uses LibSQL by default and DuckDB for observability. Local files (mastra.db, mastra.duckdb) are git-ignored.

---

Deployment

Any platform that runs Next.js 16 will work — Vercel is the path of least resistance. You will also need to:

1. Deploy the Trigger.dev worker — pnpm dlx trigger.dev deploy (requires a TRIGGER_SECRET_KEY set in the worker environment).
2. Run the Supabase migrations in your production project (both files in supabase/migrations/, in order).
3. Set every environment variable from the table above in your hosting platform. Treat SUPABASE_SECRET_KEY as server-only.

---

Contributing

Issues and PRs are welcome. Before opening a PR:

1. Run pnpm lint.
2. Run pnpm build to make sure type-checking passes.
3. If you change an agent's instructions, add a short note in your PR describing the expected behavior change.

There is no test suite wired up yet — manual testing through the dashboard is the current expectation.

---

License

This project is licensed under the Elastic License 2.0 (ELv2). See LICENSE.

In short: you can use, copy, modify, and distribute the code, but you may not provide it to third parties as a hosted or managed service that gives users access to a substantial set of its features. Internal forks and modifications are fine; spinning up a competing hosted product is not.

This is not an OSI-approved open-source license. If you need a different license for your use case, contact the project owner.

---

Acknowledgments

• Mastra for the agent framework
• Exa for semantic search
• Jina for clean page extraction
• Trigger.dev for reliable long-running tasks
• Supabase for auth and Postgres
• shadcn/ui for the component primitives
• The Hacker News community — the people posting, asking, and Show-HN-ing in those threads are the actual product.